By
After Donald Trump’s 30-point victory in the Iowa caucuses, many political observers responded by noting how the results were in line with expectations. That’s true: It was widely assumed that the former president would win by a wide margin, and he did.
But those expectations were shaped in large part by statewide polling, which proved to be quite good. Consider the results by candidate:
Donald Trump
- FiveThirtyEight’s final: 52.7%
- RealClearPolitics’ final: 52.5%
- Actual results: 51%
Nikki Haley
- FiveThirtyEight’s final: 18.7%
- RealClearPolitics’ final: 18.8%
- Actual results: 19.1%
Ron DeSantis
- FiveThirtyEight’s final: 15.8%
- RealClearPolitics’ final: 15.7%
- Actual results: 21.2%
Vivek Ramaswamy
- FiveThirtyEight’s final: 6.4%
- RealClearPolitics’ final: 6.8%
- Actual results: 7.7%
Given the circumstances, these results are pretty impressive. New York magazine’s Ed Kilgore did a nice job summarizing why pollsters faced real challenges, which they appear to have overcome:
It’s tough enough to estimate likelihood to vote in a party primary (primary polling is typically much less accurate than general-election polling). It’s much tougher to figure out who is going to be willing to attend a caucus, which requires a much greater investment of time and energy. Add in the ferocious and constantly changing weather afflicting this year’s caucuses, and you have a recipe for a polling disaster.
And yet, there was no disaster. DeSantis fared better than the surveys indicated, but in general, pollsters had a good night.
Or more to the point, they had another good night. I’m mindful of the fact that the industry has taken some hits in recent years, especially after the 2016 presidential race, but pollsters fared well in the 2018 cycle; they were pretty close in the 2020 popular vote; and polling in the 2022 cycle has been fairly described as “historically accurate“ and “great.”
Will this continue in the coming months? Time will tell. But for now, those who’ve concluded that political polling is useless, and the data is better left ignored, should probably give the industry a fresh look.